It's generally still the case. I have two workarounds. Allow replication from the cloud in the event that it is cleared out. And provide a mechanism for the user to manually backup and restore.